Skip to main content
leant reduction in the birth rate. In the Philippines and East Pakistan, the frequency of earlier child deaths in a family is a good predictor of further births. Although a reduction in child mortality may be associated for 10 or 20 years with an increased rate of population growth, this lag in the decline in birth rates may be seen as a logical consequence of the preferred pattern of family formation discussed before. Parents would appear to bear the number of children they believe they need to bear to reach a desired family size, given the death rates that once prevailed. Adjustment of reproductive behavior becomes noticeable only after the parents have reached their completed family-size goal and seek to avert further births.
As the older group of "family planning" mothers becomes substantial in a less developed country, 10 to 20 years after the reduction in child mortality, there emerges a strong statistical association between the incidence of child death and the birth rate 2 and 3 years later. Indeed in Taiwan, where the data are best designed for examining this relationship, the loss of a child less than age 15 is associated with a rise in birth rates 3 years later more than sufficient to replace the lost child and to increase the number of children likely to survive to age 15.*
The way in which birth rates adjust to the prior level and trend of death rates has been studied in detail in only two countries, Taiwan and Puerto Rico, both of which have experienced relatively low levels of child mortality for a decade or more. It is therefore suggested by some analysts that there is a threshold below which child mortality must fall before parents perceive the change and begin to seek to limit births (18). Perhaps the more modest reductions in child death rates sustained in Egypt and Colombia over the last 2 decades have not been sufficient to initiate a general decline in fertility. No completely satisfactory test of the "threshold hypothesis" has yet been performed, but data now being analyzed for low income, relatively high-mortality regions in East Pakistan provide some evidence at the family level of the anticipated compensatory relationship between child mortality and subsequent fertility.'
*This strong short-run overcompensating relationship appears not to be a manifestation of a biological feedback mechanism (based on lactation, etc.), for it is weakest among the more fertile women between the ages of 20 and 29. Rather the relationship seems to be abehaviorally determined response which is statistically strongest for women over 29, among whom the proportion practicing birth control is greatest. See (3).
• Should some form of the "threshold hypothesis" be confirmed, quite different population policy strategies might prove appropriate to different parts of the world. As a hypothetical example, Egypt and East Pakistan might find it worthwhile to concentrate social investments in health and education until changes in the parents' environment create the demand for restricting fertility and only then employ substantial resources in family planning and birth control services. Perhaps, too, modest investments in prenatal and infant nutritional supplements might be a more cost-effective route to reducing maternal and child mortality below some threshold level than waiting for the general